When printing black-and-white on variable contrast paper with a color-head enlarger, the usual tactic is to make a single exposure through a discrete filter, such as 20M (magenta) or 50Y (yellow). That specific filter is selected to afford good overall contrast to the print, which may then be fine-tuned by dodging or burning selected portions of the image. An alternative technique—so-called split-filter printing—is to expose a print twice, once through an intensely magenta filter and then through a strong yellow filter (or vice-versa). With this approach the relative times for the two exposures are adjusted to create a print of the correct overall contrast. The idea is to use a magenta exposure that provides good shadows to the print and a yellow exposure that will secure good highlights. We have seen writers claiming that this technique produces better results, but PT Editor Scott Lewis was told years ago that this latter technique produced results similar to using a filter in-between the two extremes, with no particular benef it other than more busy work. He asked us to do some comparative testing. Specifically, we were interested in three things:
• Do both approaches afford easy access to the same range of print contrasts (grade numbers)?
• If so, are the curve shapes (the path from shadow to highlight) the same with both techniques when adjusted to produce the same grade number?
• And, if so, how do exposure times compare?
We explored these issues, once again using one of the darkrooms of Edgar Praus (4photolab.com). Edgar operates a full service traditional-digital custom lab that may well be the last one remaining between Toronto and New York City.
Figure 1 examines the contrast range available when the color head is used in its traditional mode with glossy Ilford Multigrade IV RC Deluxe Paper. We made several step-tablet exposures with single filters ranging from 180Y to 180M, and measured the resulting contrasts for each, expressed here as grade numbers. The relationship is not particularly smooth, but it is clear that filtration can be identif ied that will yield any grade number from about a minus one to nearly five.
To prepare a comparable chart for split-filter printing, we exposed samples through a pair of filters, 180M and 180Y, varying the relative durations of the two exposures. There was no point in using more than 180 M or Y for the exposure pair as higher values would have only lengthened exposure times while not offering any more contrast range. It should be noted that the choice of which exposure to make first proved immaterial. In Figure 2 we illustrate the grade numbers that resulted as a function of the percentage of the total exposure time that was made through the 180M filter. For example, if a step tablet were exposed for 10 seconds through the 180M f ilter followed by 40 seconds through the 180Y, its data point would appear opposite 20% on this figure’s abscissa. The same wide range of grade numbers is again available in a smooth, rather more linear progression as the 180M portion of the combined exposure is varied from none to total. Neither technique has an advantage in the paper contrasts it affords.
But what about curve shapes? Contrast, whether measured as grade number or log exposure range, specifies only the log E distance between two widely separated points on the characteristic curve, a highlight density of 0.04 above Dmin and a shadow point at 90% of Dmax. The path between these points—the shape of the curve, and hence the tonalities of all the intermediate grays—can vary enormously while the contrast remains invariant. We know we can obtain the same grade number, anywhere between –1 and 5, with either single- filter or split-filter exposure. But might they display different curve shapes, thereby leading to prints of different appearance? Single filters typically produce a curve that looks like a smooth elongated letter “S.” We anticipated that with split filtering we might encounter a long, low-contrast toe attributable to the 180Y exposure running abruptly into a steep, high-contrast shoulder deriving from the 180M exposure. Happily, though, our worries were unfounded.
Figure 3 reproduces three pairs of curves, the members of each pair having nearly identical grade numbers. The blue curves all represent single-filter exposures and the red ones split-f ilter exposures. The three pairs were selected to illustrate rather low, medium, and rather high contrasts. The Log E axis has no meaning (beyond the fact grid lines are one stop apart) because we have tried to position the curves in an uncluttered visually instructive fashion. In particular, each red-blue pair is just slightly separated to help you appreciate our conclusion: Whether any particular contrast (grade number), is arrived at via a single-filter or a split-filter exposure, the resultant smoothly sigmoidal curve shapes are virtually indistinguishable. When contrasts are matched with the two techniques, tonalities will match at all densities and the prints will be interchangeable. Again, neither technique has an advantage.
This leaves our third issue of relative exposure times, which may well be the Achilles’ heel of split-filter printing. We need to select a print density at which to make comparisons, and our choice is that of a diffuse highlight, 0.04 above Dmin. This choice is based on the premise that diffuse highlights are the least negotiable part of an image: They must always be just off-white, never dirty gray or empty white. We will adjust and note the relative exposure times needed to force all of the curves to overlap at this density. And to dispense with the word “relative,” we will arbitrarily assign an exposure time of 10 seconds to the white light exposure, which, lacking any filtration, is necessarily the shortest time of all the exposures.
Figure 4 displays these corresponding times as a function of their respective grade numbers. The blue data points represent the single-filter approach. The shortest exposure (10 seconds) is with white light (grade number 0.6). As yellow or magenta f iltration is increased, printing times lengthen: six-fold at the extreme of 180Y and ten-fold at 180M. But the red data points, documenting total exposure times with split filtering, reveal a signif icant drawback to this approach: All of these times are intermediate between the longest values corresponding to 180Y and 180M. This means for all of the most commonly used grade numbers, exposures are much longer with split f iltering. Add to this the need to readjust filter and timer settings between the two exposures and the result is considerably more time in the darkroom.
Our conclusion? Split-filter printing with a color head seems like the long way around the barn. It does not afford access to a print appearance (contrast, curve shape) unattainable with single filters; it just takes longer to get there. It could prove useful if printing with boxed filters or an Ilford head, which only increment contrast in half-grade steps. In this case split filtering would allow finer tuning of print contrast. There is also a tactic wherein a print receives a basic overall single-filter exposure and then selected portions of the image are burned or dodged through an extreme filter. This to be sure can be a useful technique, but it is a fundamentally different from what we are here terming split-filter printing.
Finally, a footnote to those who may want to explore split-filter printing as we have described it here: All of the numbers and data points we have presented are unique to the paper (identity, batch, age), and choice and condition of the processing chemistry and enlarger we used. Change any of these and—while the patterns, trends and conclusions we have documented would be unaltered—specific numeric values almost certainly will change from darkroom to darkroom; i.e., personal testing is a must.